Ability to clearly identify folders as model references

Feature Requests
1

Working with AI can be like being Tom Cruise in The Edge of Tomorrow. Constantly needing to walk through the same steps to get to where you were in the last conversation.

Pain Point
Often when I need to iterate, I’ll try to use a previous version of my product to point on things that have worked before. “Look at this file and see how we did it before when it worked”

It works in non-editor environments but it continues to fail terribly for me - because the AI model will often write updates to the reference project and not the active project.

And when it does, I often can’t get the context out of the model.

Feature request: Lock in some files as reference, as clearly labeled for model learning and not writing. File names would be similar so it needs an object definition of sorts.

2

Thanks for the suggestion Skifree - I totally get the Edge of Tomorrow reference, and the team is already looking into solutions for this. We know how frustrating it can be when the AI writes to reference files, and we’re working on ways to make the distinction between reference and active files much clearer